Factors to Consider in Recording Avian Sounds

1 Factors to Consider in Recording Avian Sounds D A V I D C. W I C K S T R O M

I. Introduction II. Terminology III. Microphones A. Types of Converters and Their Characteristics B. Polar Patterns C. Microphones for Special Applications D. Microphone Specifications E. Accessories F. Collecting Specific Sounds in the Field G. Some Illustrative Tests H. Additional Considerations in Evaluating Microphones IV. The Tape Recorder A. Some Fundamentals and the Direct Recorder B.The Audio Recorder: A Specialized Direct Recorder C. Other Types of Recorders D. Metering E. Recording Formats and Tape Recorder Performance F. Some Tape Recorder Problems G. Tape Recorder Maintenance V. The Magnetic Tape A. The Tape and the Tape Recorder B.Tape Storage and Handling Problems VI. Signal Monitoring and Modification A. Monitoring the Recording B. Signal Modification: Filters, Equalizers, Limiters, and Noise Reducers VII. The Recording System A. Interconnection of the Components B. Batteries C. Selecting a System: A Sample Problem

ACOUSTIC COMMUNICATION IN BIRDS VOLUME 1

2 3 4 5 6 7 7 14 15 17 27 29 29 30 31 36 38 40 41 41 41 42 44 44 45 47 47 47 48

Copyright © 1982 by Academic Press. Inc. All rights of reproduction in any form reserved. ISBN 0-12-426801-3

2

David C. Wickstrom VIII. Summary References

I.

50 51

INTRODUCTION

The recording of bird sound was once a scientific achievement in itself. Today, using relatively inexpensive equipment, anyone can make subjectively good recordings. While this may be a pleasurable pastime, it can scarcely be called a scientific endeavor. Documentation and calibration are two key words for the serious recordist. The scientific utility of a given recording is diminished greatly if good data are not kept on what is being recorded. Similarly, if the characteristics of the recording system are not documented, the precision of conclusions drawn from analysis of the recordings is severely degraded. An ideal recording system would reproduce in the minutest detail an exact electrical analog of the sound present at the time of recording. Unfortunately, this ideal can only be approximated. Most recording equipment design is predicated on its use for human speech or music. These assumptions are not necessarily valid for avian recording. If recor­ dists are not aware of the limitations of their recording systems, they can un­ knowingly precondition their data. In the case of field recording where the expense of getting to the location can be substantial, it is illogical to use anything but the best equipment that can be accommodated. Even if the recording is for a specific research project, it will, ideally, be deposited with an archive. When a recording becomes a library resource, the better the quality and documentation, the wider its usefulness. From the moment sound emanates from a bird, it becomes contaminated. Contamination comes from the environment and the recording equipment. The finished recording consists not only of the bird sound, but of all the sounds present in the environment at the time of the recording as well as artifacts generated by the recording equipment. A primary objective in recording is to optimize the amount of desired sound with respect to the amount of extraneous information, in other words, to maxi­ mize the signal-to-noise ratio. To accomplish this requires an understanding of the behavior of sound in air (see Chapter 5, this volume), the behavior of the subject, and the characteristics of the recording system. No matter what tech­ nological wonders are available for postrecording processing, the easiest time to maximize the signal-to-noise ratio is when the recording is made. There is no piece of electronic apparatus that does not, in some way, change the signal passing through it. In most components, the action of the device is largely beneficial (the amplifier increases the signal), but there is some cost

3

1. Recording Avian Sounds

(addition of noise and distortion). It is important to know what types of distortion a system introduces since the usefulness of the final recording is affected. It is noteworthy that the existing standards for the measurement of distortion do not correlate well with perceived quality. In this chapter I will first review some terms, and then break the recording system into its component parts: microphones, recorders, tape, and signal-pro­ cessing equipment and interface components. The final part of the chapter will discuss the combination of components into a recording system.

II.

TERMINOLOGY

Decibels: One of the most common terms used when working with sound is the decibel (dB). An understanding of the term is crucially important when working with sound. Two things to keep in mind are that the term dB (1) simply means that a logarithmic power ratio is being used and (2) unless referenced to something, does not represent an absolute measurement. For power the formula is:

dB = 10 logg For voltage: dB = 20 l o g | i

where E is voltage (see Davis and Davis, 1975). Since it is possible to find sounds occurring over a range of 180 dB (although 130 dB is more usual), and most recording systems have a dynamic range of around 50 dB, one can see the necessity of making some informed decisions. Various standards exist specifying the methods to be used when evaluating, audio and sound equipment. Some of the organizations involved are the Ameri­ can National Standards Institute (ANSI), the International Organization for Stan­ dardization (ISO), the International Electrotechnical Commission (IEC), the So­ ciety of Motion Picture and Television Engineers (SMPTE), the Institute of High Fidelity (IHF), and the Audio Engineering Society (AES). Other than these organizations, manufacturers of audio test equipment are excellent sources of information and application notes. The following descriptions are purposefully general as space does not permit delving into the various standards. When comparing equipment be sure that the same test methods were used. Frequency Response: This is the degree of amplitude change with respect to frequency. This specification is usually presented graphically with amplitude on the v axis and frequency on the x. The ideal of no change in amplitude with frequency would be presented as a straight line, i.e., a "flat" response. In written form, upper and lower frequency limits are given along with the limits of

4

David C. Wickstrom

amplitude variation. It is important to have a system response as wide as the frequency range of the subject. Signal-to-Noise Ratio: Expressed in decibels, the signal-to-noise ratio is a measure of the distance between a reference level (usually the maximum ca­ pability) and the noise floor of the device. In many cases the specification is "weighted" according to a specific frequency curve. Ideally, an unweighted specification is also given. The unweighted value is the more useful number for analysis as it is not a frequency-sensitive specification. Dynamic Range: Related to the signal-to-noise ratio, the dynamic range gives the range of amplitude variation that a device can accommodate at a given setting. Noise: Sometimes a specification simply labeled " n o i s e " is given. This is a measurement of the device's inherent noise level. The reference point and the band width of this measurement should be given somewhere in the specifications (e.g., all measurements referenced to 0.775 V). Phase Shift: This specification is seldom seen. If predictable phase response is a requirement, the system should be checked by a competent technician. Drift, Wow, Flutter, Scrape Flutter: These specifications refer to tape record­ ers. Drift is usually considered to be a speed variation of up to 0.1 Hz.; Wow from 0.1 to 10 Hz; Flutter from 10 to 300 Hz.; and Scrape flutter from 3 to 5 kHz (McClurg, 1976). The specification most often given is the combined Wow and Flutter weighted by some standard curve. Total Harmonic Distortion (THD): This specification gives the level of har­ monic content at the output of a device, relative to the level of the fundamental. It is most often expressed as a percentage but it can also be expressed in dB. For example, harmonic content 40 dB below the fundamental is the same as 1% THD. Intermodulation Distortion (IM): The purpose of this test family is to ascertain how much interaction there will be between two tones passing through a device. IM is expressed as a percentage. The appropriate standard should be consulted for the specifics of the test. It is worth noting that the two distortion specifications refer to a device being tested under static operating conditions. At the present time, dynamic tests are evolving but are yet to be standardized.

III.

MICROPHONES

The first component in the system that sound encounters is the microphone, which has the crucial task of converting some aspect of acoustical energy to an electrical analog. The description of a microphone includes the type of converter, its polar pattern, frequency response, and efficiency.

1. Recording Avian Sounds

5

A. T y p e s of Converters and Their Characteristics 1. The Dynamic

Microphone

The dynamic microphone consists of a diaphragm which is attached to a tube. Around the tube is wound a coil of wire. The coil is positioned in a magnetic field in such a way that, when sound moves the diaphragm, the coil moves through the magnetic flux, inducing a current in the coil. In simplest form, the dynamic microphone will exhibit a broad midrange peak in its frequency response (Olson, 1957). Flatter amplitude response is achieved by distributing the mass of the diaphragm and by using filter networks (usually acoustic, but sometimes electrical or combinations). The specific virtues of dynamic microphones are their durability, environmental immunity, and electri­ cal simplicity. The inherent noise of the dynamic is limited to that generated by thermal agitation of the coil and diaphragm and is therefore very low. On the negative side, the dynamic, with its large diaphragm and voice coil, is susceptible to mechanical noise. The average output level of the dynamic is lower than a condenser. If the following electronics are lacking in gain or are noisy, the system noise may be quite high. The high mass of the diaphragm assembly makes it difficult for the dynamic to handle fast rise time signals. The dynamic is not a good choice for extremely low frequency response and its stability is affected by temperature. 2. Condenser

Microphones

Condenser or capacitor microphones use varying capacitance to generate the electrical analog. The diaphragm is made of very thin conductive material stretched over a support. Close behind the diaphragm is a conductive plate. The two surfaces form a capacitor. As the sound moves the diaphragm, the capaci­ tance varies. This variation of capacitance is used in one of two ways to create the audio output. The most common method is to polarize the element using either an externally supplied dc voltage, or an electret material (a plastic that has a virtually permanent charge). The capacitance change causes a proportional change in voltage that, after an impedance change, becomes the audio output. The other method places the capacitor microphone element in the resonant circuit of an oscillator. The oscillator changes frequency with the change in capacitance, creating an FM output. The FM is demodulated to produce the audio output. The condenser microphone once confined to studio use is a type that no longer lends itself to generalizations. With the advent of solid-state electronics and electret elements, its usefulness has been greatly expanded. The condenser can offer wide, dependable frequency response, high output, long-term stability, and relative insensitivity to mechanical noise. While not always the case, the design can offer predictable phase response and fast rise times. The classic difficulties with condensers involve the need for associated elec-

6

David C. Wickstrom

tronics. With one version, the requirement is for a polarizing voltage and an impedance converter. The electret eliminates the need for external polarization, but still requires the impedance converter. The impedance converter is com­ monly called either an amplifier or preamplifier, although I know of no condens­ er microphone with gain in its electronics. The FM microphone requires an oscillator and demodulator. Assuming the electronics of a condenser microphone to be trouble free, power is still required to operate them. The light diaphragm, desirable from the sonic standpoint, is easily contami­ nated. The greatest enemy of the condenser is humidity. If moisture finds its way into the element, it will raise the noise of the microphone, and in extreme cases cause conduction through the element, resulting in large amounts of low-fre­ quency noise at the output. With the electret, the charge on the element will be diminished, reducing the sensitivity of the microphone. In addition, the electret's sensitivity is permanently reduced by elevating its temperature above that for which it was designed (usually above 100°-120°F). 3. The Piezoelectric

Microphone

The piezoelectric microphone, otherwise called crystal or ceramic, uses the piezoelectric effect to produce its output. A piezoelectric material is by definition any material that produces an electrical voltage when deformed. A diaphragm is either attached to this material or the material's surface itself becomes the di­ aphragm. While attractively simple, this microphone type has not kept pace with others in quality, and is found only in specialized applications. 4. The Ribbon

Microphone

The ribbon microphone uses a very light metallic ribbon suspended in a strong magnetic field. As the ribbon moves, a voltage is developed proportional to the particle velocity of the sound wave. B. Polar Patterns The polar pattern of a microphone is a graphic representation of the sensitivity of the microphone with respect to the angle of incidence of the sound. There are three basic patterns: omnidirectional, bidirectional or figure eight, and unidirectional/cardioid. Various prefixes, such as super or hyper, are added to better define the degree and type of pattern. Representative polar patterns are shown in Fig. 1. It is easy to forget that a polar plot represents a three-dimensional pattern, as shown in Fig. 2. (This figure assumes symmetry, which is not always the strict case.) In their elemental form, all microphones except the ribbon have an omnidirec­ tional pickup pattern. All except the ribbon respond to variation in air pressure. In the case of a single diaphragm microphone, acoustical networks are used to create the desired pattern. The purpose of the delay network is to cause the sound

7

1. Recording Avian Sounds

©9

e A

B

C

D

E

Fig. 1. Microphone polar patterns. (A) Omnidirectional. (B) Cardioid. (C) Hypercardioid. (D) Supercardioid. (E) Shotgun.

coming from the undesired direction to arrive at both sides of the diaphragm at the same instant, resulting in a net pressure differential of zero. The difficulty of creating a microphone-sized acoustical delay equally effective at all frequencies is one reason for the frequency dependence of polar patterns. As the frequency rises, the wavelength diminishes to a point where the physical size of the micro­ phone exerts control over the pattern. With the addition of directional sensitivity, the microphone no longer responds to the pressure but to the pressure gradient. C. Microphones for Special Applications 1. Measurement

Microphones

Most microphones sold are manufactured for use in standard audio applica­ tions. There are a few manufacturers that offer or specialize in measurement grade microphones. The best known of these is Bruel and Kjaer. The primary purpose of these microphones is to make calibrated acoustic measurements. Extensive documentation is offered as well as an excellent series of application notes. If reliable, repeatable readings are required, the usefulness of this type of equipment should be explored. 2.

Hydrophones

A hydrophone is a microphone designed for use underwater. Under certain conditions they can be used in air. Specific literature should be consulted. 3. Wireless

Microphone

A wireless microphone substitutes a radio link for the microphone cable. Commercial units exist with excellent specifications. It should be noted that some incorporate noise-reduction systems. The manufacturer should be con­ sulted for specific applications. D . Microphone Specifications It is necessary to view published specifications with a degree of skepticism. They apply only to new units and even then the production variation can be quite wide. If there is a specific parameter that are important to a recording, the

8

David C. Wickstrom

Fig. 2. Three-dimensional microphone polar patterns. (A) Cardioid. (B) Omnidirec­ tional. (C) Figure eight. (D) Shotgun. (E) Supercardioid (see pp. 9-12). (Courtesy Sennheiser Corp.)

Fig. 2 (continued)

David C. Wickstrom

10

Fig. 2 (continued)

Fig. 2 (continued)

David C. Wickstrom

Fig. 2 (continued)

13

1. Recording Avian Sounds

microphone should be tested. If stability and consistency of the system's perfor­ mance are important to a recording project, means of calibrating the system in the field must be obtained. 1. Output

Level and

Sensitivity

For avian recording, the efficiency of a microphone is of major importance, since the sound level is in many cases extremely low. Even the best electronics contribute some noise. All things being equal, however, the microphone with the higher output level is the more desirable. Two terms used to describe a micro­ phone's efficiency in converting sound to electricity are output level and sen­ sitivity. When comparing microphones, be sure to read the specification care­ fully and ascertain what reference is being used. There are two reference levels in common use. One is a level of 74 dB sound pressure level (SPL) at the micro­ phone. The other is 1 Pascal or 94 dB SPL. Before the specifications can be compared, one must know to what the reading is referenced and whether it is an open circuit or a power reading. If the specifications describe a microphone operating into a matching load, the output will be 6 dB lower than in the unloaded configuration, which is the way most microphone input circuits are designed. It is possible to reconcile the various rating methods so the output levels can be compared (see Section VII,C). 2.

Noise

Noise is specified only for microphones that have associated electronics. A noise specification of 17 dB means that the noise at its output is the same as if the microphone had no inherent noise, and was in a sound field of 17 dB SPL. Again, this is usually a weighted specification and the appropriate standard should be consulted. The noise of a microphone is sometimes given as a signal-to-noise ratio. In most cases, the reference point is 94 dB SPL. This is all well and good except that in field recording it is not uncommon to have an SPL at the microphone of 50 dB. What was an impressive signal-to-noise ratio of 78 dB in reference to 94 dB SPL becomes 34 dB in reference to the 50 dB SPL level. In many cases the noise of a condenser microphone is determined by replacing the capsule with a capaci­ tor of the same value. This does give the noise of the electronics, but it assumes a perfectly dry element, which is seldom found in the field. 3.

Impedance

The output impedance of a microphone is specified in ohms. For most micro­ phones that are called low impedance, it is on the order of 200 ohms. The importance of this specification is that, to achieve the best signal-to-noise ratio,

14

David C. Wickstrom

the impedance of the microphone should be appropriate to the input of the recorder. 4. Maximum

Level

The maximum sound level a microphone can handle is specified in decibels SPL for a given level of distortion. If the microphone is used close to the source it would be a good idea to consult a reference such as Beranek (1971) on the behavior of sound and microphones in the near field. 5.

Polarity

The polarity of a microphone is usually defined as the number of the in-phase connector pin. This means a positive-going pressure at the microphone produces a positive-going voltage on the specified pin. 6.

Powering

A condenser microphone needs some source of electricity to operate. This can be supplied internally by a battery enclosed in the microphone housing or exter­ nally by a power source. External power sources can use mains current, bat­ teries, or an associated piece of equipment as the source of electricity. If the power source is external to the microphone, there must be a way of delivering electricity to the microphone. This can be done with additional con­ ductors in the microphone cable or by combining the power with the audio, using one of several different methods (Phantom, T, etc.). Each method has its relative merits. When buying a condenser microphone make sure that its powering re­ quirements have been accounted for.

E. Accessories The selection of a microphone involves more than just the initial selection. The microphone chosen must be easy to handle and protected from potential damage. By looking through the catalogs, one can find various useful accessories for mounting microphones. 1.

Windscreens

A major problem in field recording is eliminating the noise and subsequent overloading caused by wind striking the diaphragm. Wind manifests itself as low-frequency noise that may not be evident on the meter or monitor. The effectiveness of windscreens is seldom given, and in most cases, the recordist must test their effectiveness. The larger the windscreen, and the higher its acous­ tic resistivity, the better it will reduce wind noise. A windscreen is always a

1. Recording Avian Sounds

15

compromise between the degree of shielding offered and the degree to which it changes the response of the microphone. Most windscreens sold today are made of a plastic foam, either permanently attached to the microphone or designed to be slipped on. Two common pitfalls with this type of windscreen are the use of a foam not designed for the purpose, and the failure to cover the back vents of a directional microphone. Some microphones have a built-in windscreen for use as a blast filter. Their primary purpose is for the close miking of vocalists, but they are not always effective as field microphones. In extreme circumstances, two windscreens can be com­ bined, the resulting loss of response being the lesser of the evils. 2.

Filters

If a high-pass filter is included in a microphone, it can be an aid in reducing wind and handling noise. Its amplitude and phase response should be such that it does not interfere with the sound being recorded and, as in the case of any signalmodifying device, its use should be noted. 3. Shock

Mounts

Noise generated by the movement of the microphone can come from a number of sources. It can result from the microphone shaking, the cable conducting noise to the microphone housing, and either the microphone or the cable brushing against objects. Various shock mounts, cables, and handles are sold to aid in reducing handling noise, and some succeed. It is also possible to buy micro­ phones with internal shock mount systems. Any shock mount will be sensitive to the weight of the microphone, the frequency of vibration, the rate of accelera­ tion, and the type of attachment point. One should try the mount, making sure it is workable as a piece of field gear, and that it actually attenuates handling noise. If a shock mount seems to be doing little or no good, make sure the cable (or anything else) is not short-circuiting the mount. Many manufacturers sell shock cables and proper mechanical terminations along with their mounts.

F. Collecting Specific S o u n d s in the Field 1. The

Parabola

A parabola focuses onto a single point incoming sound waves that are parallel to its axis. Its effectiveness is determined by the diameter of the reflector in relation to the wavelength of the sound. Its gain and directivity increase propor­ tionately with decreasing wavelength. For wavelengths larger than the parabola's diameter, the response is predominantly that of the microphone itself. For a parabola to be minimally effective at 100 Hz, the diameter must be a little over

16

David C. Wickstrom

11 ft. The practical solution is to use as large a reflector as possible in conjunc­ tion with an omnidirectional microphone. The resulting combination is om­ nidirectional up to the point the diameter becomes significant with respect to wavelength and has increasing directivity and gain as the frequency increases. Approximate wavelengths of sound in air corresponding to common parabola diameters are: 36 inches = 370 Hz; 24 inches = 565 Hz; 18 inches = 750 Hz; 13 inches = 1040 Hz. Placement of the microphone in the parabola is governed by the designer's choice of focal point. A parabola can be constructed with varying degrees of curvature, each equally effective as a reflector. The deeper the curvature, the closer to the back of the dish is the focal point. If the focal point is placed inside the plane of the edge of the reflector, it is shielded from the wind, but since the reflector is also a resonant cavity, the microphone picks up the resonance. Noise from the movement of the parabola (either from wind or handling) will be louder the closer the microphone is to the parabola. If the focal point is outside the plane of the edge of the reflector, these effects are minimized but the benefit of shielding from the wind is lost. Commercial realizations exist with the focal point at any of these locations and a number exist with the focus at the approxi­ mate edge. The microphone is not always placed at the precise focal point. This is done both to broaden the beam width at the extremely high frequencies and to involve more of the microphone's diaphragm. Note that not all microphones work efficiently in a parabola and a gain test should be a part of choosing a microphone. 2. Shotgun

Microphones

The shotgun microphone is a cardioid microphone fitted with an interference tube. The interference tube causes phase cancellation of sound arriving off-axis. A phrase often heard describing a shotgun microphone is that the microphone has "greater reach." This does not mean that the microphone has some ability to collect more energy from the source. As the distance from the source increases, the energy per unit area decreases. The shotgun microphone can respond only to the energy reaching its diaphragm. The advantage lies in its ability to reject more of the off-axis sound. Consider that the microphone is surrounded by a sphere of noise. Interpose in this sphere of noise the three-dimensional polar pattern of Fig. 2. It can be seen that the more narrow the pattern, the less noise will be accepted. The wider the microphone's pattern, beyond that needed to receive the desired sound, the more noise is accepted. If the noise is not equally distributed spatially, then the microphone's specific angular rejection characteristics become important. Since it is not always possible to aim a microphone precisely, some manufacturers construct their shotgun microphones in a way that reduces the high-frequency beaming. This is similar in effect to the slight defocusing of a parabola.

17

1. Recording Avian Sounds

I

Fig. 3.

Polar response of Dan Gibson parabola. (A) 500 Hz. (B) 2500 Hz. (C) 5000 Hz.

G. S o m e Illustrative Tests 1. Representative

Equipment

To examine how these devices function, I tested three different parabolas as well as one omnidirectional microphone and one shotgun microphone. These tests were performed under controlled conditions and, while comparable, they are not absolute. The Dan Gibson parabolic microphone is an assembly incorporating an 18inch reflector, dynamic microphone, electrical equalizer, and monitor amplifier. Its polar response is shown in Fig. 3, its axial frequency response in Fig. 4. Since the response curve includes the response of the filter network, the filter's re-

00

QP 0124

Multiply Freq. Scale by

Fig. 4.

Zero Level:

(1612/2112)

Axial frequency response of Dan Gibson parabola, M setting.

A B C Lin:

Bruel & KjcBr

Potentiometer Range:

50

HR

Rectifier:

Lower Lim.

FRAG-

20

W _APJL mm/sec y r Speed: H

Paper Speed:_JJL mm/sec 0n75

Measuring Obj' 'dBudB

IdBIdB 8-60

Dan Gibson elect.roni.C30i response

it

+ >> r

-U *^!> —L-

Rec. No.: Date: Sign.: QP 0 1 2 4

^ 5

0L0

i //

1

I"

0

-I

Multiply Freq. Scale by . Fig. 5.

Zero Level:.

—I—l—n—\~ 6^45

M \ "ft—

4—

—j—

- f r ­

I

itu J&_ (1612/2112)

Electronic response of Dan Gibson parabola amplifier, M and V settings.

4^30

! I

! I

! S ! 1 J I

-

2H5

—!—1—1—1—I •

1 1 1 1

A B C Lin.

0J0

20

David C. Wickstrom

sponse alone is shown in Fig. 5. The assembly offers the advantage of a translu­ cent dish to allow for visual aiming. It does not offer the option of interchange­ able microphones, and it requires batteries for operation. The Sony PBR-330 has a diameter of approximately 13 inches. It is offered as an accessory item, and is designed to accommodate a wide range of micro­ phones. The parabola was fitted with a Sennheiser MKH-104 omnidirectional condenser microphone. The polar pattern and axial frequency response are shown in Figs. 6 and 7. At the same time, the microphone's axial response was tested without the reflector (Fig. 8). Band-limited pink noise was fed through the source (200-8000 Hz, 18 dB/octave) and the output level of the microphone was noted. The difference between the microphone alone and with the reflector is

Fig. 6. Polar response of Sony PBR-330 parabola with Sennheiser MKH-104 micro­ phone. (A) 500 Hz. (B) 2500 Hz. (C) 5000 Hz.

Brllel & KjcBr

50r25

Potentiometer Range:

Measuring Obj.:

Sony PBR-330 w/ MKH 104 -IQdB

^Q—dB Rectifier:

RMS

Lower Lim Freq.:

20

z r

Speed:mm/sec

W H

Paper Speed: JJL

mm/sec

30

Rac. No.: Dale: Sign.: QP 0124

or.

10 20—'Hz 50 Multiply Freq. Scale by

10000

20000 40000 D A B C Lin. (1612/2112) A B C Lin.

Fig. 7. Axial frequency response of Sony PBR-330 parabola with Sennheiser MKH-104 microphone.

Bruel & Kicer

Potentiometer Range:

dB

RMS

Rectifier:

Lnwar

Mm

Frpg

•

20

h?

Wr.

Spgfld:

25

mm/sec

Paper Speed:_LQ_ m m / s e c

50 Measuring Objdo

10n75

40

IdBIdB

Sennheiser ) MKH 104

8-60

30}

6^ 45

sn.69977 4^30 Rec. No.:

10

2H

Date: Sign.: QP 0124

10 20 Hz 50 Multiply Freq. Scale by

Fig. 8.

100

200

Hz

500 1000 Zero Level:

0_ 2000

Hz

5000

10000

20000 40000 D A B C Lin. (1612/2112) A B C Lin.

Axial frequency response of Sennheiser MKH-104 microphone.

23

1. Recording Avian Sounds

+14.9 dB. The Sony offers the advantage of a plastic reflector, and also gives the user the option of selecting microphones appropriate to particular applications. In that both of the parabolas discussed are relatively small for many avian sounds, the same tests were conducted using a 36-inch aluminum parabola fitted with the Sennheiser MKH-104. The results are shown in Figs. 9 and 10. While its additional size is a hindrance in the field, its better gain and directivity indicate it should be considered. The difference between the microphone's out­ put by itself and mounted in the reflector is + 1 8 . 9 dB. For purposes of comparison, the response of a Sennheiser MKH-805 shotgun microphone is shown in Figs. 11 and 12. I

Fig. 9. Polar response of 36-inch parabola with Sennheiser MKH-104 microphone. (A) 500 Hz. (B) 2500 Hz. (C) 5000 Hz.

RS Bruel & KjcBr Measuring Obj.

56 inch

Potentiometer Range:. dBtdB

50 .dB Rectifier:.

M Lower Lim. Frag :

0

2 H?

Wr.

Speed:_J^_ mm/sec Paper Speed:_12_ mm/sec

4 0

IdBOdB 8-60

|

diameter parabola -5QdB w/MKH-104 Rec. No.:

i

6^45

3Q

20

4^30

10'

2-1

Dale: Sign.: 200 Q P 0124

Multiply Freq. Scale by .

Fig. 10.

Hz

500 1000 Zero Level:

2000

Hz

5000

10000

20000 40000 D A B C Lin. . (1612/2112) A B C Lin.

Axial frequency response of 36-inch parabola with Sennheiser MKH-104 microphone.

25

1. Recording Avian Sounds

I

Fig. 11. Polar response of Sennheiser MKH-805 shotgun microphone. (A) 500 Hz. (B) 2500 Hz. (C) 5000 Hz.

2. Positioning

a Microphone

in a Reflective

Environment

To illustrate one of the problems in microphone positioning, consider Fig. 13. It can be seen that in one position there are many potential paths by which the sound can reach the microphone. In the other, where the microphone is placed at a boundary, there is only one path (until very short wavelengths are involved). An experiment was set up with the microphones positioned as indicated. The resultant response curves are shown in Fig. 14. It can be inferred that, in a reflective environment, microphone placement is important if anomalies in the recording are to be avoided.

BrUel & Kioer

Potentiometer Range:

50

dB

Rectifier

RMS

Lower Lim

Freq:

20

H7

Wr.

Speed:

25

mm/sec

Paper Speed:

LQ_ mm/sec

50 Measuring Objdo _Sennheiser40|

MKH 805 s n . 2668

i

3

Q

Rec. No.:

Date: Sign.: QP 0124

10 20 Hz 50 Multiply Freq. Scale by

Fig. 12.

10000

20000 40000 D A B C Lin. (1612/2112) A B C Lin.

Axial frequency response of Sennheiser MKH-805 shotgun microphone.

1. Recording Avian Sounds

Fig. 13.

27

Two microphone positions relative to a reflective surface.

H. Additional Considerations in Evaluating Microphones I. Suitability

of a Microphone

for an

Application

When evaluating a microphone, its physical suitability for the intended use must be considered as much as its electroacoustic capabilities. When procuring equipment of any kind, it is important to consider what it was designed to do. This may be stating the obvious, but if a piece of equipment is designed for a specific purpose, it is likely to be best suited to that application. Since equipment is seldom designed for use in avian recording, one is required to choose devices that have applications analogous to avian recording. By comparing the design goals with your application, you can get some indication of suitability. In most bird sound recordings, I would say that the signal-to-noise ratio is fixed at the time it is made. In other words, the environmental noise recorded exceeds the electronic noise of the system. To improve the signal-to-noise ratio requires that more signal reach the microphone element. Four ways to accom­ plish this are: (1) get the source and microphone closer; (2) capture more of the energy; (3) reject unwanted sound; (4) some combination of these. Getting the microphone closer to the source is an attractive option. While it requires a lot of wire, proper impedance microphones, and much patience, it is a technique that should be used more than it is. The need to capture more energy indicates the need for a parabolic reflector. Rejecting unwanted noise is the purpose of all directional microphones, especially the shotgun microphone. 2. Checking a Microphone

in the Field

If one is trying to make recordings where absolute SPL information is re­ quired, an appropriate microphone with a compatible pistonphone or acoustic

Bruel & KicBr

5X1

Potentiometer Range:

RMS

dB Rectifier:

Lower

t im

Frer^ •

20

h?

Wr.

Speed:_6*L_

mm/sec

Paper

50r25i

Measuring Obj.jJU 30ft. from .source 30 microphone 4 f t . above ' •surface Rec. No.:

Dale:

Speed:_JJL_ mm/sec ion75 8 60 IdBIdB ~ 6-45

2 0

4^30

0

1

2-15

Sign.: QP

0124

10 20 Hz 50 Multiply Freq. Scale by

100

200

Hz

500 1000 Zero Level:

2000

Hz

5000

10000

20000 40000 D A B C Lin. (1612/2112) A B C Lin.

J Bruel & Kjosr

Potentiometer Range:

_JLQ__dB

Rectifier:

RMS

Lower Lim

Freq :

20

H7

Wr.

Speed:

63

mm/sec

Paper Speed:

Measuring Obj.-

LQ_ mm/sec

0q75

IdBIdB 8H60

24 5

QP

0124

10 20 Hz 50 Multiply Freq. Scale by

Fig. 14.

200

Hz

500 1000 Zero Level:

2000

Hz

5000

10000

20000 40000 D A B C Lin. (1612/2112) A B C Lin.

Frequency response of Sennheiser MKH-104 microphone positioned as in Fig. 13.

1. Recording Avian Sounds

29

calibrator should be carried (manufacturers: General Radio, Bruel and Kjaer). For a rough check of microphone deterioration, the following techniques can be used. When the microphone is new and assumed to be operating correctly, make a recording of your own voice outdoors in a quiet nonreflective environ­ ment. Position yourself at a good working distance from the microphone. Note the settings of all the controls and your distance from the microphone. Save this recording and periodically repeat the test. By comparing the recordings, you can hear if something has changed. Of course, if a new microphone of the same model as the one in question is available, a simple A - B comparison can be performed. For a rough indication of polar response, you can walk a circle around the microphone and note your position on the tape. By listening to this recording, you can hear how angular incidence affects the recording. Any crucial parameter should be verified by further testing.

IV.

THE TAPE

RECORDER

A. S o m e Fundamentals and the Direct Recorder There are four basic elements involved in recording and playing back magnetic tape. 1. The transport mechanism which moves and guides the tape. 2. The recording section which takes the signal from the microphone and records it on the tape. 3. The playback system. 4. The tape itself. Magnetic recording is possible because certain materials can be magnetized by an external force and remain in that state after the force is removed. To make a recording, the magnetic material is moved past a transducer or " h e a d " at a linear rate. The tape and the record head form a magnetic circuit. As the tape is moved past the head, the magnetic particles are magnetized in proportion to the current flowing in the record head at that instant. One of the bothersome characteristics of the magnetic medium is that it does not acquire magnetism in direct proportion to the applied magnetic force. To overcome this problem, an ultrasonic signal, or bias, is added to the record signal to linearize the recording characteristic. The frequency chosen for the bias must be high enough so that it will not be recorded and not beat with harmonics in the signal. In addition, each magnetic material has its own magnetization curve. This means in practice that the tape and the recorder must be matched. If one is not sure about a particular machine/tape match, a qualified technician should be consulted.

30

David C. Wickstrom

To play back the signal, the tape is moved past another transducer similar to the record head (combined in some machines). As the tape is moved across the gap of the play head, the magnetic force causes an electrical signal to be pro­ duced. The level of the output produced is dependent on the rate of change across the gap. In other words, unless the tape moves, no signal is produced. As the output is produced by the rate of change of the signal, it can be seen that it must vary with frequency. If the bandwidth of interest is recorded on the tape at an equal level of magnetization and then played back, the output will rise with frequency at a rate of 6 dB per octave. In any tape recorder, compromises exist between tape speed, gap width, highfrequency limit, and system noise. If the head gap is decreased, the high-fre­ quency performance improves, but the low-frequency output decreases. As the tape speed is increased, the recorded wavelength of a specific frequency in­ creases, but more tape is used. When the recorded wavelength reaches the same dimension as the gap width of the play head, there can be no change, hence no output. If the rudimentary recorder just described were actually constructed, it would be called a "direct" recorder. The advantages of direct recording are simplicity as well as appropriateness to a wide variety of applications. Depending on tape speed, it is possible to obtain frequency response of at least 5 Hz to 500 kHz. A dynamic range of 50 dB is not unusual. The major disadvantages of direct recording are limited low-frequency re­ sponse, amplitude variation, and time variation. Limited low-frequency response is not a problem for most avian recording as most machines have dependable response to 30 Hz. Time variation, both short and long term, can be a problem if the work requires accurate frequency determination or a low residual FM. Amplitude variation, or modulation deserves a little more discussion. Most often called dropout, the amplitude modulation (AM) in analog recording is caused by oxide variation and/or by the tape moving away from the head. The tape can be lifted from the head in a number of ways. The two most common are surface imperfections in the tape itself and surface contamination of either the tape or the heads. The effect can be quantified by the equation: Playback Drop in dB = 54 DIL, where D equals the separation from the head, and L equals the wavelength. For example, if the tape is lifted from the head 0.0005 inch (human hair diameter, approximately 0.003 inch), the drop in level of a 10-kHz signal recorded at 7.5 inches/sec is 36 dB. B. The A u d i o Recorder: A Specialized Direct Recorder In order to minimize noise and use the maximum capacity of the tape for audio recording, circuits to add preemphasis and deemphasis to the signal are added to the basic direct recorder. Early on it was observed that in speech and orchestral

1. Recording Avian Sounds

31

music, the predominant energy is in the midband, decreasing at the extremes. Using this spectral distribution as justification, the signal is equalized to better distribute the energy. The level is increased with frequency during recording and attenuated during playback. The record equalization (preemphasis) is not the perfect inverse of the playback in that various record/play losses are addressed at the same time. These excess losses are greater at slower tape speeds. Because of the preemphasis of the record signal and tape oxide characteristics, the saturation point for an audio recorder is not linear with frequency. In other words, the higher the frequency of a signal at a given input level, the closer the recorder is to severe distortion. At tape speeds of 15 inches/sec and above, which require less pre-emphasis, this effect is almost nonexistent. At lower speeds the amount of record equalization is substantial. This is particularly important to the avian recordist, because virtually no devices used to monitor levels are spectrally weighted and the spectral distribution of avian sound bears little resemblance to the speech and music curves used to justify the choice of the standard record/ playback equalization curves. The result is that, given a bird whose song has its predominant energy in the 4to 6-kHz band, and a tape recorder running at 3.75 inches/sec, an unweighted meter reading could easily be in error by 10 dB. In this example, the actual recording level, due to the action of the record equalization, will be 10 dB higher than the level indicated by the meter. If the level is set to the meter's maximum, the actual record level will be well into the distortion range of most recorders. If the tape speed is slower, or the sound higher in frequency, the error will be greater. Figure 15 is a group of amplitude-versus-frequency charts made on one channel of a Nagra IV-S operating at different speeds, the input level being decreased in 10-dB increments. Figure 16 gives the results of the same tests using a Marantz PMD-220 cassette recorder. The loss of high-frequency re­ sponse at the higher levels is due to saturation.

C. Other T y p e s of Recorders 1. Instrumentation

Recorders

The commercial realization of the simple direct recorder is available as one version of an instrumentation recorder, or a "scientific recorder." If one is recording a signal with a spectral distribution totally incompatible with an audio recorder, an instrumentation type of recorder should be considered. Instrumentation recorders offer many options. Care should be taken during selection to make sure the correct options are purchased for the type of recording intended. Tapes do not necessarily interchange among different brands of instru­ mentation recorders, a capability we assume with audio recorders. Along with direct recording capability, instrumentation recorders can offer

•

w

'<*<>*'>•>''

* \u«ra I \ - S ' Record/ J'CSPQDGO

ff

<•

.5^'

id

R<'

**

RMS

;u

,

1 -

"

,

-

. i o n

,

^

^ ^ ^ ^ ^ ^ ^ ^ ^ ~ ' '

y ^ - ^ ^ ^

j

*

,

S

"

"

"

~ •

;

GPC24

>/.

C

~

L

-

1

*

7 . .

„

_

-

-

A B O .

ft C

::: 1 ::: ::

w

" -,

response

; ...;

5

•

-

7\, 5 JLn/sec

QP 0124

:

; s •

•

;

(/ Frjq

-

'

^-^^.^x^^-^^^^.^

Swi « 0 /

*

"~^*NX

. I

Mo.' ;

!

_

„

'

\ .

Z ' o l ^ i

.

M

//^t^*

* IH
; «

^ B C bn

A Br,r4

c 4 K v.-

Q P 0124

P-A

• ~,v>P,>"

V

Multiply Freq. Scale by

.

.50. dB Rpc» *«.r ........ RMS

:

_

!/ ^r

;

r

Zero l e v e l : . ™

,20

W»

_

>

.100

-

'V "

11612/21121

,

'

5

A

8

C Lm.

Fig. 15. Record-playback response of a Nagra IV-S tape recorder (left channel) at 0, - 1 0 , - 2 0 , and - 3 0 dB (Nagra IV-S meter). (A) 15 ips. (B) 7.5 ips. (C) 3.75 ips.

33

n BrUei & KjoBr

Potentiometer Range:—50

dB Rectifier:

BMS__ Lower

Lim. Freq.:

20

H? Wr. S p e e d : _ 6 3 _ mm/sec

Paper

Speed:__3L_

mm/sec

Measuring Obj.

QP

0124

10 20 Hz 50 Multiply Freq. Scale by

200

Hz

500 1000 Zero Level:.

2000

Hz

5000

10000

20000 40000 D A B C Lin. (1612/2112) A B C Lin.

Fig. 16. Record-playback response of a Marantz PBR-220 cassette recorder operating at 1.875 inches/sec at 0, - 1 0 , - 2 0 , and - 3 0 dB (Marantz meter).

1. Recording Avian Sounds

35

FM and digital recording. FM recording differs from direct in that a carrier is modulated by the input signal. It is this modulated carrier that is recorded. This in turn is demodulated upon playback. FM recording evolved in answer to the two basic limitations of the direct process, the inability to record low frequencies (to dc) and amplitude instability. One of the limiting factors of the FM system is the ability of the tape transport to maintain absolute tape speed since any variation in speed will manifest itself as spurious modulation and increase the noise. The FM system preserves the phase versus frequency characteristic of the input signal. The system is more compli­ cated electronically and is far less efficient in its utilization of tape. Although it gives extremely low frequency response, its high-frequency capability is gener­ ally limited to about one-fifth that of direct recording at the same tape speed. The FM instrumentation recorder is not a general purpose recorder, but if one needs to record extremely low frequencies, have faithful phase-versus-frequency response and good immunity to amplitude variations, it may have to be considered. 2. Digital

Recorders

Digital recording uses the magnetic medium to store logic states. This type of recording is used extensively in the data processing industry, to a degree in instrumentation recorders, and is beginning to find application in audio record­ ing. Since the signal consists of binary information the recording process itself is relatively simple. To record sound, the signal must be converted to digital information. This is called an analog-to-digital (A-to-D) conversion. Once digitized, the digital bit stream is recorded. When played back it must be reconverted to an analog signal. This process is called digital-to-analog (D-to-A) conversion. The A-to-D converter takes discrete samples of the analog waveform (typ­ ically 50,000 per second for commercial audio recorders) and quantizes them. The more often we sample, the wider the bandwidth of the system. In addition, the more bits used to quantify the sample, the more accurately the signal is represented, and the lower the noise. At the moment, systems in common use have a sampling rate of approximately 50 kHz and use 12- to 16-bit quantization. To protect the recording from lost information, various error-correcting schemes have evolved. Discussing the relative merits of these is beyond the scope of this chapter. Suffice it to say, the better the error-correcting scheme, the more storage space it requires. Digital audio recording requires compromises between bandwidth (or sam­ pling rate), accuracy of each sample, signal-to-noise ratio, effectiveness of the error-correcting system, and amount of tape required. At the moment, there are no standards for these variables, with resulting problems of tape interchangeability. Given more time for the technology to evolve, digital audio recording will

36

David C. Wickstrom

become a method in general use with capabilities for accurate recording beyond those available with most analog recording systems.

D . Metering 1. Types of

Meters

No matter which method of recording is chosen, the machine will have a specific limit to the range of signals it can handle. To get the best signal-to-noise ratio one records at as high a level as possible without reaching the point at which the signal becomes grossly distorted. To do this some means of indicating the level of the signal, usually some form of meter, is required. The most familiar is the Volume Unit (VU) Meter. A 1-dB level change of a sine wave equals a change of 1 VU. This meter was developed to answer the young audio industry's need for a standard operating level. Previous to this, each organization had its own stan­ dards and program interchange was difficult. In 1939, the characteristics of the VU meter were standardized along with a reference level of 1 mW into 600 ohms. The section of the standard that is particularly important to avian record­ ing is the one specifying meter ballistics. The standard requires the meter to give a 100% reading with a pointer overshoot of not less than 1% and not more than 1.5% (0.15 dB). Furthermore, the pointer must reach 99% of the 0-VU mark in 300 msec (Tremaine, 1969) in response to the sudden application of a sine wave at a level equivalent to 0 VU. It is a testimony to the design that after 40 years, the VU meter is still an accepted standard. The meter, however, was designed to indicate levels of human speech and orchestral music. Peaks in sound can occur in much less than 300 msec, making it possible for a VU meter to give an inaccurate reading of the true level. In answer to this potential deficiency, various peak-indicating meters have evolved. One standard for peak metering is the Peak Program Meter (PPM). Its standard calls for an integration time of 12 msec (Gordon and Wood, 1979). If the fallback time were the same as the integration time as in the case of the VU meter whose pointer returns to minimum scale in 300 msec, the pointer of the PPM would be a blur for most program material and the human eye would find it impossible to register the reading. To overcome this problem, a longer fallback time is specified. For the PPM it is 1 sec. There are numerous metering schemes of varying efficacy. It is unfortunate 4 expensive that their integration characteristics are not always specified. In tless equipment, it is common to find a meter that, although labeled V U , " does not 4 meet the standard. A key word in all meter specifications is 'program." The intent of the designers is not to provide an absolute measure of the minutest peak but to provide a means to easily set levels during a program. The predictable and

37

1. Recording Avian Sounds

periodic nature of most musical and spoken program material is what allows the VU meter to be the useful tool it is. In the case of bird sound where it is possible to find peaks in the 5-msec range occurring over relatively long intervals, it is possible to have substantial metering errors. 2. Metering

Errors

There are two interlocked variables that cause metering errors. One is the duration of the pulse itself. The other is the time between succeeding pulses. In the case of the VU meter, the analogy of pushing a swing is appropriate. The swing will move a given distance either in response to one mighty shove or a succession of small shoves. The longer between shoves the less effective they are. I carried out an experiment on three different meters using a continuous tone set to yield a reading of zero. The mode of the oscillator was then changed to generate a pulse of sine waves of varying length. Interpulse distance (the time between pulses) was also varied. Tables I and II show the metering error for various settings of the oscillator. If the pulse is short enough and/or the interpulse time is long enough substantial metering errors can result. With any metering system, it is necessary to determine what a reading means relative to a given amount of distortion of the recorded signal. In the case of a recorder equipped with a VU meter, the zero calibration point cannot correspond to the saturation point since any signal occurring at a higher level for less than 300 msec will distort. The designer must decide how much headroom to leave between the maximum reading on the meter, and the point at which objectionable distortion occurs. Meter specifications of some current recorders are: Nagra IV-S, meter integraTABLE I Response of Three Meters to a Constant Amplitude 10 kHz Tone Burst of Varying Duration" 50-msec Repetition rate Burst length (msec)

Nagra

Uher

100.0 50.0 10.0 5.0 1.0 0.5

0* 0 0 -7.5 -12

0'' 0 0 -10 -20

VU

Nagra

-11 -15 OS< OS

0 dB 0 -1 -4 -16 -20

a b Tone burst repeated every 50 and 500 msec. c Continuous time. Off scale.

500-msec Repetition rate Uher 0 dB 0 -4 -8 c OS OS

VU - 5 dB -10 -19 OS' OS OS

38

David C. Wickstrom TABLE II Response of Three Meters to a 0.1-msec Signal Repeated at Different Intervals" Interpulse interval (msec)

Nagra

Uher

VU

0.05 0.10 0.50 1.00 5.00 10.00 50.00 100.00

0 dB 0 0 0 -6 -9 -19 OS

0 dB 0 0 0 -5 -11 OS OS

- 3 dB -5 -12.5 -16 OS'' OS OS OS

11 b Amplitude

set to 0 dB for a continuous 10-kHz signal.

Off scale.

tion time 10 msec plus or minus 20%, 1% third harmonic distortion at 15 ips at + 4 dB; Ampex ATR-101 studio recorder, 300-msec integration time (VU meter), less than 0.3% third harmonic distortion of a 1-kHz tone at 0 VU, less than 3% at + 9 VU; Uher 4000 Report Monitor, meter integration time 30 msec, decay time 400 msec, 1% distortion at 0 on meter. The Uher's meter is equalized to give an indication of the effect of preemphasis. This is an excellent idea and it is unfortunate that it is not offered on more recorders. A case can be made for the superiority of each of the above metering systems. What is more important, though, is to be aware of which metering system a recorder uses, and how close its 0 mark is to the point of objectionable distortion. Armed with this information, and a good monitor system, one can begin to evolve an operating procedure appropriate to a recorder. In view of the variables involved in metering sounds, it should be apparent that the widespread custom among avian sound recordists of arbitrarily adjusting recording input levels to —10 or less on whatever meter their recorder is equipped with will not produce optimum recordings. E. Recording Formats and Tape Recorder Performance Figure 17 illustrates common track widths in current use. The terms full, half, and quarter track are used, but the actual track widths are not fully indicated by these terms. For this domestic manufacturer, the half track and quarter track use only 32% (not 50%) and 18% (not 25%), respectively, of the width of a full track on '/4-inch tape. The casette uses narrower tape (or 0.150 inch) and one track of a stereo casette is only 10% of the width of a full track recording on !/4-inch tape.

0.027" 0.1456

r

_i STEREO PAIR 0.0235 T R A C K S ON 0.0355 C E N T E R S PAIRS SPACED ON 0.0866 C E N T E R S STEREO

0.060 T R A C K S ON 0.087 CENTERS OPPOSITE D I R E C T I O N S MONO

CASSETTE 0 . 1 5 0 " TAPE (0.150/0.144)

1

F U L L TRACK

0.075 T R A C K S ON 0.159 CENTERS C/2 T R A C K USES A 2 T R A C K H E A D ) / TRACK AND 2 TRACK

2

0.043 T R A C K ON 0.136 CENTERS 0.043 T R A C K S ON IN ONE H E A D S T A C K 0.068 C E N T E R S TAPE T R A C K S I D E N T I C A L % TRACK AND 4 TRACK % " TAPE (0.244/0.248)

Fig. 17. Common tape recorder formats and track widths. (Courtesy Ampex Corp.)

40

David C. Wickstrom

These percentage differences result from the necessity of having a guard band between the tracks to prevent one track from bleeding into the other. This bleed always occurs to some degree. Most manufacturers specify the amount of bleed, called crosstalk. A way to compare various formats is to consider the recording area per unit of time. One second of information at 7.5 ips full track uses a recorded area of 0.234 x 7.5 inches or 1.755 sq. inches. A cassette recorder, using one stereo track with a width of 0.0235 inch and running at a speed of 1.875 ips, will use a recording area of 0.044 sq. inch. The casette track uses about 2.5% of the area used by a full track recorder at 7.5 ips. Tape contamination that would obliterate the entire casette track would leave 97.5% of the information intact on a full track recording. The choice of format affects the signal-to-noise ratio. The smaller the track width, the more the signal-to-noise ratio will deteriorate. A stereo machine has two tracks available at a time for recording. A common field practice is to feed the signal from the single microphone to the two tracks thereby doubling the recording area. What is gained is not double the usable playback area but redundancy. It is virtually impossible to sum the two outputs of a stereo recorder without distorting the signal. This results from what are vir­ tually normal head alignment problems in most recorders. For example, the wavelength of a 15-kHz signal at 1.875 ips is 0.000125 inch. If one track is offset from the other half of that amount or 0.0000625 inch, there will be a phase difference of 180° at 15 kHz. When the outputs are summed, there will be complete cancellation at that frequency. In practice, the phase difference be­ tween the tracks always varies slightly because of machine instability and/or tape variation. The result of summing the two tracks is to create a complex filter, the action of which is greater at high frequencies and the characteristics of which vary as the tape is moved across the head. F. S o m e Tape Recorder Problems Speed instability while the recorder is being moved can be a serious problem for the field recordist. Many designers assume that a portable tape recorder will not be in motion while it is being operated. The easiest way to check a machine for speed instability is to play a tape of a continuous tone. Move the machine about and listen for pitch variations. Some recorders run off speed or run at different speeds under different condi­ tions of temperature and battery capacity. The practice of recording a tone from a pitch pipe or tuning fork, common in the early days of field recording, is still a recommended procedure. The tone is recorded at the beginning of each recording and checked upon playback. Speed variation will manifest itself as frequency modulation of the signal. Be careful about attributing FM to the subject without first checking the recorder for flutter occurring at the same rate.

41

1. Recording Avian Sounds

Audio recorders produce a high-frequency amplitude variance (Budelman, 1978). The program material itself will cause the high-frequency output to vary. It can amount to an amplitude change of 2 - 4 dB, and it increases as the tape speed gets slower. Functionally it is program-dependent AM. G. Tape Recorder Maintenance Maintenance is crucial for proper recorder performance. Routine maintenance should be performed at least every 10 hr of running time. Routine maintenance must include, as a minimum, a thorough cleaning and demagnetization of the tape path. If the proper test and calibration equipment are not available for verification of the recorder's performance, a qualified technician should be found and a maintenance schedule set. For machines in daily use, this verification should be performed at least once a week. If the machine is consistently meeting specifications at the 1-week interval, it may be possible to extend the service interval. Equipment seldom fails catastrophically, and gradual deterioration may go unnoticed. If you are going to an area where it is impossible to obtain normal service, carry manuals and wiring diagrams for the equipment. If service is required, and a technician can be found, having this information is an invaluable aid. Where possible carry spares for failure-prone parts. Be aware of the maintenance requirements of equipment and be realistic when purchasing. One tape recorder requires a series of screws to be removed to allow head cleaning and another uses heads that are out in the open. The easier a machine is to maintain, the greater the likelihood that it will be maintained.

V.

THE MAGNETIC

TAPE

The quality of the final recording is determined not only by the recorder but by the tape itself. There are three basic components to recording tape: the base film, the binder, and the magnetic material. The most common magnetic material is a type of iron oxide. The base film is usually a polyester material that has been prestretched to increase tensile strength. Until recently, cellulose acetate was in common use as a backing material. Paper has also been used as a base and many other materials have been tried. A. The Tape and the Tape Recorder 1. Matching

the Tape to the

Recorder

The interface between the magnetic medium (tape) and the tape recorder is in every way a very sensitive one. Recording tape is the result of the skillful blending of many variables into a product with optimum operating parameters for

42

David C. Wickstrom

a given application. Since needs vary, there are many different types available. The magnetic properties that are of most interest are bias requirement, output capability, and high-frequency sensitivity. To realize maximum performance from a tape, the recorder's bias level, record level, and record equalization must be adjusted to each tape stock's specific properties. While there is some interchangeability among different tape stocks, one should be very careful. The recorder's performance should be verified by mea­ surement after a change in tape. Good frequency response alone is not an indica­ tor of correct adjustment. Because both bias level and record equalization will cause the high-frequency output to vary, it is possible to obtain a relatively flat frequency response through compensating errors in the two settings. Since bias level also affects tape output and distortion, flat frequency response may have been achieved at the expense of other parameters. Inaccurate bias settings cause increased noise and distortion. 2. Contact

of Tape with Recorder

Head

Intimate contact between the head and tape is achieved in two basic ways. One uses a pressure pad to press the tape against the head. Although used in some reel-to-reel machines this is most commonly found in cassettes where the pres­ sure pad is an integral part of the cassette itself. Alternatively, the tape transport is designed to put the tape under tension and to pull the tape against the head(s). If the pressure (or tension) is too high, excessive tape and head wear occurs. If it is too low there is excessive amplitude variation. For maximum high-frequency output and accurate interchangeability, the tape must be held so that the gap in the head is perfectly perpendicular to the edge of the tape. In reel-to-reel recorders the guides that accomplish this are held to close tolerances. In the cassette recorder the cassette itself supplies some of the tape guiding. As a result the mechanical quality of the cassette has a strong effect on overall recorder performance. B. Tape Storage and H a n d l i n g Problems As magnetic recording is a relatively new process, we do not have substantial data on the effects of long-term storage. However, sufficient time has elapsed to indicate definite trends. The magnetic material itself is not likely to deteriorate. However, its magnetic state can be changed by an external field, destroying or modifying the recorded material. Many problems of tape storage and handling can be traced to a failure or deformation of the base material or a failure or degradation of the binder. Most short-term tape deterioration is related to deformation of the tape. The most common causes are: too rapid winding, uneven winding, improper tape tension, bent reels, mixing tape stocks on a single reel, and poorly slit tape (recording

1. Recording Avian Sounds

43

tape is cut from wide rolls and sometimes the finished width varies). While dust does not attack tape directly, it can cause loss of tape to head contact and can scratch the tape or head. For this reason, areas where tape is handled should be clean. In the field, recorder covers should be kept closed whenever possible. Print-through occurs on all recordings. It is caused by the signal on the tape being recorded onto the layers of tape directly adjacent to it and is heard as a preceding and following echo when the tape is played back. If there is an external magnetic field present, or the temperature is high, print-through becomes worse. The distance between layers is fixed by the thickness of the tape. This is one of the arguments against thinner tape. There are tapes on the market that are specifi­ cally formulated to reduce print-through and can be effective if used properly. Magnetic fields can cause effects ranging from increased noise levels to complete erasure. Temperature extremes can deteriorate the base. It is common practice to store tape on the take-up reel as it comes from the machine after being played (tail out). This practice has many benefits. Most machines do their best job of tape spooling (evenness of wrap) at playback speed, and the tape tension is more consistent than when spooled under fast wind conditions. An additional benefit of tailout storage is that fast rewinding before playback can reduce print-through as much as 10 dB. Figure 18 shows a level/ time representation of print-through. The second graph is of the same tape played just after rewinding. Notice that both pre- and postprint have been reduced. For short-term storage, avoid extremes of environment and rapid changes of temperature and humidity. For long-term archival storage, professionals should be consulted. For polyester base type, the general rule is the cooler and drier the better. For acetate base type, the humidity should be around 50% since as the backing dries it loses its plasticisers, and the tape becomes brittle, cups, and

44

David C. Wickstrom

cracks. On the other hand, acetate tapes stored in this fashion (high humidity) can be subject to fungus attacks. Polyester should be dry as the binder is least active at low humidity levels. Long-term storage of magnetic tape requires a substantial commitment to climate control, beyond the range of the average recordist. If possible, the original or a safety copy of valuable field and research recordings should be archived. Many archivists recommend periodic rewinding and respooling of the tape at 6-month to 1-year intervals. This reduces print-through and balances uneven tension which may have built up within the tape pack from temperature or humidity changes. After use, the tape is played off the reel, the end affixed, the recording acclimatized to the storage environment, placed in a plastic bag prior to being placed in its container, and then stored on edge in the archive. One of the advantages of reel-to-reel tape is that it can be edited by cutting and splicing desired portions. Until recently, most splicing tape had an adhesive which flowed with temperature and eventually dried out. If at all possible, an unspliced copy should be archived as well. Cassettes do not usually contain tape stock optimized for long-term storage. In addition, they cannot be mechanically edited. A cassette may be archived, but a high-quality reel-to-reel copy should be made and stored as well.

VI. SIGNAL MONITORING AND MODIFICATION

A. Monitoring the Recording 1. The Sense of Hearing

and the Skill of

Listening

The most critical elements in any monitor system are our ears. Good hearing by itself is not sufficient, however, and needs to be used in concert with other techniques. A monaural recording does not supply the information necessary for our ears to locate a sound. Many recordings are not as good as they might have been because the recordist ignored extraneous sounds in the environment and fixated on the sound of interest. This is an ability that the recording system does not have. The effect is even more pronounced when there are corresponding visual cues to distract us. The best recordings are made by people who remain aware of the total sound environment whenever they are recording in the field. Another problem is that our hearing fatigues and acuity degrades during long listening sessions. Our ears, while remarkably resistant to abuse, are not im­ mune. If subjected to loud sounds, the ears' sensitivity changes. If this is tempo­ rary, it is called temporary threshold shift. If the sound is loud enough, the threshold shift and other hearing damage can be permanent. Protect your hear-

1. Recording Avian Sounds

45

ing! If you are around loud sounds—e.g., power tools, firearms—use hearing protectors. Be aware that there is a time/intensity relationship in inducing hearing loss; in other words, a softer sound for a long time may accumulate as much damage as a louder sound for a short time. Seek the advice of a professional to obtain devices to protect your hearing as some on the market have little or no effect (cotton in your ears offers almost no protection). If you rely on your hearing in your work it is especially desirable to have an annual audiometric exam. Listening is a skill that must be developed and must not be confused with hearing. Hearing is a sense. Listening is an active endeavor and requires the use of one's intelligence. In 1960, Dr. P. P. Kellogg described a simple experiment to illustrate how little of an unfamiliar sound we perceive. A simple but convincing experiment is to record your own voice, speaking your own name or some simple phrase or sentence. Reproduce these sounds in reverse. Then diligently try to reproduce with your voice what you hear. Record this jumble of sounds, and in turn, reproduce them in reverse. Normally what you hear will be astonishing, and an indication of how superficially you comprehend the intracacies of unfamiliar sounds. Usually great improve­ ments are achieved with practice. It is much the same with vocalizations of birds and other animals. With experience we become sensitive to smaller and smaller variations.

Kellogg assumed that the recorder being used was full track, reel-to-reel. 2. Headphones

and

Speakers

For critical evaluation of a recording, it is necessary to present our ears with the best possible reproduction of the sound. A monitor system can use either headphones or a loudspeaker as the output transducer. Whichever is used, the quality is important. The loudspeaker built into the tape recorder is completely inadequate forjudging the quality of the recording. For field recording, a good pair of headphones, properly matched to the recorder, is a must. For evaluating and editing, a good loudspeaker system is preferred. While not usually requiring maintenance as frequently as the rest of the recording system, headphones and loudspeakers are not immune to damage or drift and should be tested periodicaliyB. Signal Modification: Filters, Equalizers, Limiters, and N o i s e Reducers If one intends to manipulate a signal with various processing devices, it is necessary to understand how they operate. The following descriptions are intended as an introduction. Before using these devices, it is recommended that not only the manufacturer's literature and instruction manual be consulted, but that a broader perspective of their operating principles be acquired by reading appropriate literature (e.g., Tremaine, 1969; Woram, 1976).

46

David C. Wickstrom

A wide range of signal-modifying devices is available. Filters and equalizers can be used to modify the shape of the spectrum and minimize extraneous sounds. Band-pass filters allow a desired portion of the spectrum to pass. Bandreject filters reject a portion of the spectrum. High-pass filters allow the portion of the spectrum above their setting to pass through. Low-pass filters do the opposite. Graphic equalizers are arranged so that the settings of their controls give an approximate graphic indication of how the amplitude response is being changed. A graphic equalizer is usually made up of filters of either full octave or one-third octave width. These filters can be arranged in such a way that when used together a combined response curve results. This is not always the case and if combining action is desired it should be verified by testing. The parametric equalizers are so named because they have controls that allow every parameter of the equalizer to be varied. The center frequency is continu­ ously adjustable as well as the bandwidth. The degree of gain or attenuation is also adjustable. If one intends to use equalization in the dubbing or analyzing of a recording, it is worthwhile to examine the available literature. It is very easy to cause distortion of the signal with an equalizer and the system should be tested carefully. Limiters and compressors are special purpose audio devices used to control automatically the dynamic range of the signal. They are sometimes found on tape recorders under the guise of automatic level controls. As a rule their use is not appropriate in field recording. If there is a special application that demands automatic control of level, the literature on these devices should be consulted. Noise-reduction systems can be divided into two groups: encode-decode sys­ tems, and single-pass systems. The two common encode-decode systems are D o l b y tmand D B X t .mThese two share the common goal of reducing tape noise and increasing the dynamic range. To accomplish this, they require that the signal be encoded during the recording process. Reciprocal processing is re­ quired for playback. Discussing the relative merits of these systems is beyond the scope of this paper, but it should be mentioned that, although both systems succeed in reducing noise, they do so at some expense to overall quality. If the application is at all critical and the use of noise reduction is anticipated, the characteristics of the system should be examined closely and its performance verified. It is this author's experience that some bird sounds fool all the systems some of the time. Single-pass noise-reduction systems do not require encoding of the signal; rather, they act on the signal as it is played back. With the exception of comput­ erized noise reduction, the single-pass systems use the signal itself to control one or more filters so arranged as to attenuate the portion of the spectrum unoccupied by the desired program. Effectiveness is governed by the filters themselves and the machine's ability to differentiate the desired signal from noise. From an archival standpoint, all noise reduction systems represent a modification of the

1. Recording Avian Sounds

47

recording, and their use should be noted. Single-pass systems, while sometimes helpful for a specific project, are not appropriate for general use. By striving for the best possible recording in the first place, the necessity of postprocessing of the recording can be avoided.

VII. THE RECORDING SYSTEM When a recording of a bird is made it is done with a recording system, not a group of modules. It is possible to assemble a system of excellent components that will produce abysmal results. Each component can have acceptable perfor­ mance, but the combination may accumulate errors in a way that makes the final result unacceptable. The recording system should be tested and calibrated to make sure its errors are not significant relative to the sensitivity and type of the intended analysis.

A. Interconnection of the C o m p o n e n t s Space does not permit a thorough discussion of interconnection techniques. Impedance, balancing, wire and connector types are all important. If electrical impedance and balanced and unbalanced lines are unfamiliar terms, a good reference such as Davis and Davis (1975) should be consulted. Wire and connec­ tors present mechanical as well as electrical difficulties. Wire needs to be elec­ trically correct for its application as well as durable and flexible at working temperatures. My preference is for a system comprised of as few connections as possible. Those that are necessary should be of the same type. This minimizes the need for repair and makes it easier to carry spares. B. Batteries For field recording, power is generally supplied by batteries. Make sure the types used by your equipment are available in the area where you are recording. If the availability of replacement batteries is in doubt it is a good idea to carry a supply sufficient for the duration of the project. Batteries are generally classified by the elements they use to produce elec­ tricity. Currently available types include: carbon-zinc, alkaline, nick­ el-cadmium, lead-acid, silver oxide, and mercury oxide. The carbon-zinc battery is a low cost battery. Its output voltage falls gradually through its useful life. It is best used at room temperatures since its life decreases at high temperatures and its output falls at low temperatures, virtually stopping at 0°F ( - 1 8 ° C ) . An alkaline battery costs more than a carbon-zinc but it has more available

48

David C Wickstrom

energy. The output voltage falls during its life, but at a slower rate. It works well at low temperatures, and is not unduly sensitive to high temperatures. This is the best battery for most field recording. Nickel-cadmium and lead-acid batteries are types designed to be recharged. The ^ n i - c a d " is relatively insensitive to temperature although it should not be be charged at temperatures below freezing and does not store well at high temperatures. One charge will not last as long as the life of a comparable nonrechargeable type. The lead-acid battery is relatively economical if a rechargeable battery that can deliver large amounts of current is needed and size and weight are not a problem. High temperatures cause little problem, but output drops substantially at low temperatures. A car battery is a special type of lead-acid battery. Its life is shortened considerably when discharged to the limit of its capacity. The type of lead-acid battery that is built to withstand full discharges without damage is called a deep cycle battery. Silver oxide batteries are used primarily in applications where the voltage needs to be constant throughout the battery life. They are generally available only in miniature sizes offering small current capability. They are not rechargea­ ble. Mercury batteries offer good efficiency in proportion to size, and have almost constant output voltage over their life. Their high temperature capability is good but their output is nil at freezing. They are quite expensive and are used where small size is important. The "button" batteries used in some microphones are mercury batteries. They cannot be recharged. C. Selecting a System: A Sample Problem As an example of problems encountered in choosing components for a field recording system, one potential system is examined. As a recorder, the system will use a Nagra IV-S, a stereo machine. For the purposes of discussion only one channel will be considered. The recorder has been equipped with a case to protect it. A copy of pertinent sections of the manual have been placed in the cover. The heads and pinch roller are out in the open permitting easy access for cleaning and demagnetizing. The stock recorder cover accommodates only 5inch reels, and the cover should be closed whenever possible to reduce the chance of tape or machine contamination. The correct tape has been ordered on 5-inch reels. The recorder has been aligned, biased, and equalized for the intended tape stock. The Nagra manual gives the meter integration time of 10 msec, which should catch most of the peaks in the sounds to be recorded. The next problem is to select a microphone. If a humid working environment is anticipated, a dynamic shotgun offers the advantages of ruggedness and immu-

49

1. Recording Avian Sounds

nity to moisture. A possible candidate is the Electro-Voice DL-42. Its frequency response (50-12,000 Hz) is not exceptional and it will have some trouble track­ ing fast rise time sounds. However, a condenser would be trouble prone in the anticipated environment. The DL-42 comes with a windscreen and a handle, although the catalog does not show how the handle fits. The specifications of the DL-42 give its output as —50 dB, but —50 dB relative to what? In the front of the E.V. catalog, it is stated that the reference level is 0 dB = 1 mW, delivered to a load impedance equal to the microphone's impedance, with a sound pressure level of 94 dB. The test frequency is 250 Hz. In other words, the microphone was presented with a 250-Hz test tone at the level of 94 dB SPL at the microphone. The output of the microphone was measured and referenced to 1 mW. The frequency range of our intended subjects is between 1 and 6 kHz. The microphone's response is adequate at those frequencies; however, the second harmonic of 6 kHz will be attenuated. This limitation is not important for the intended analysis. The microphone's specified impedance is 150 ohms. There­ fore, it will work with the recorder's 200-ohm input position. The 200-ohm input has a specified sensitivity of 0.28 mV for a 0-dB reading on the meter with the gain control at maximum. To determine if the microphone has sufficient sensitivity when used with this recorder, it is necessary to reconcile the ratings. By current convention the microphone's impedance is not actually matched by the low impedance inputs of the recorder. Even though the input is labeled 200 ohms, the actual input imped­ ance is on the order of ten times the stated value. Such an input is called a bridging input. A power reading is converted to an open circuit voltage reading by the formula

d mB £ = V ( 0 . 0 0 1 Z) 1 0 K

+/ 160 ) l

where E is voltage, Z is impedance, and dBm is decibels referenced to 1 mW. For the DL-42, this formula gives the open circuit voltage as 2.44 mV. The difference in dB between 2.44 and 0.28 mV is 20 log (2.44/0.28) = 18.8 dB. Subtracting the 18.80 from the 94 dB required for the microphone to produce the 2.44 mV, it can be seen that it would require a SPL of a little over 75 dB at the microphone to produce a 0-dB reading on the meter with the gain control at maximum. This is slightly less sensitivity than is desired. Further examination of the Nagra manual indicates that the machine has a high sensitivity position that gives another 6 dB of gain. This extra gain reduces the SPL requirement at the microphone to 69 dB. Since more sensitivity would be desirable for a worst-case situation, a Senn­ heiser MKH-816 condenser microphone is also examined. Its sensitivity is spec­ ified as 40 mV/Pascal. A Pascal is equivalent to 94 dB SPL. Conveniently both

50

David C. Wickstrom

microphones were specified with the same SPL applied. The DL-42's output is 2.44 mV unterminated in comparison to the Sennheiser 40 mV, better than 24 dB difference. The Sennheiser is also available with the necessary windscreens and handle. Because it is a condenser, the MKH-816 requires power to operate; this can be provided by the Nagra. The Nagra's input with T-standard powering has a sensitivity of 4.2 mV, which is 23.5 dB less sensitive than the 200-ohm input. Using the recorder in this configuration, there is no significant difference in usable sensitivity between the two microphones. The Nagra's 48-V phantom input has a sensitivity of 1.4 mV, and the microphone is offered in that version. That gives an additional 9.5 dB of gain. If it were possible to power the microphone separately from the Nagra, the 200-ohm input could be used for maximum sensitivity. This can be done using the in-line battery supply available for the MKH-816. With the battery supply in the line between the microphone and the 200-ohm input, the system has the maximum possible sensitivity with this microphone. Unfortunately, the micro­ phone has a certain amount of electronic noise that will be amplified as well. In this configuration it will take approximately 50 dB SPL at the microphone to produce a meter reading of 0 dB. Unfortunately, the 1% distortion point for the recorder's 200-ohm input is specified as 54 mV. Since the microphone produces 40 mV at 94 dB SPL, it can be seen that any loud sound (above 9 dB SPL) will drive the recorder's input into distortion. This exercise illustrates the need to be aware of the specific requirements for recording the intended subject. The above examples are not intended to recom­ mend equipment, but to illustrate a thought process. This is the first step in determining how a given combination of components will work as a system.

VIII. SUMMARY The serious avian sound recordist, someone who is recording bird sound for scientific or professional use, must possess adequate knowledge, not only of the bird itself, but of the behavior of sound and the capabilities and limitations of the recording system components and the system as a whole. The concepts and the reasons for record preemphasis and the design philoso­ phy of metering systems are particularly relevant to the avian recordist. A recording that is not documented and is recorded on an uncalibrated system has limited usefulness. One that is properly made becomes a valuable research resource and should be archived. The choice of equipment appropriate to the recording of avian sounds is especially difficult in that virtually all available equipment has been deliberately

1. Recording Avian Sounds

51

designed for other purposes. While it is important to choose components care­ fully for their specific merits, it is essential that their combination result in an optimally functioning system. There is no shortcut to this knowledge. It must be gleaned from a variety of sources. The " s t a n d a r d " references, while helpful, are quickly outdated and sometimes in error. The " c o o k b o o k " approach, while of use to an amateur, can only be a basic introduction for the beginning professional. Recordists must be aware of the requirements of their intended analysis and view the limitations of their recording system in that light. Lacking thorough verification and calibration of the recording system, the analysis is likely to display not only avian sound but recording system artifact.

ACKNOWLEDGMENTS I would like to thank James L. Gulledge for encouragement and expert assistance in preparation of this text. Saul Mineroff kindly made available some of the equipment discussed.

REFERENCES Beranek, L. L., ed. (1971). "Noise and Vibration Control." McGraw-Hill, New York. Bruel and Kjaer (1977). "Condenser Microphones and Microphone Preamplifiers, Theory and Application Handbook." Bruel and Kjaer, Naerum, Denmark. Budelman, G. A. (1978). High frequency variance: a program dependent signal mode in analog magnetic tape recording. Prepr. No. 1377(E-2). Audio Eng. S o c , New York. Burroughs, L. (1974). "Microphones: Design and Application." Sagamore, Plainview, New York. Davis, D., and Davis, C. (1975). "Sound System Engineering." Sams & Bobbs-Merrill, New York. "Eveready" Battery Application and Engineering Data (1971). Union Carbide Corp., New York. Fisher, J. B. (1977). "Wildlife Sound Recording." Pelham Books, London. Gordon, J. K., and Wood, J. B. (1979). Bridging the gap between the vu meter and ppm. Prepr. No. 1518(D-7). Audio Eng. S o c , New York. Greenewalt, C. H. (1968). "Bird Song: Acoustics and Physiology." Smithsonian Inst. Press, Washington, D.C. Gulledge, J. L. (1977). Recording bird sounds. Living Bird 15, 183-204. Hassall, J. R., and Zaveri, K. (1979). "Acoustic Noise Measurement." Bruel and Kjaer, Naerum, Denmark. Hetrich, W. L. (1976). The accu-peak level indicator. Prepr. No. 1125(B-4). Audio Eng. Soc, New York. Kellogg, P. P. (1960). Considerations and techniques in recording sound for bio-acoustics studies. In "Animal Sounds and Communication" (W. E. Lanyon and W. N. Tavolga, eds.), Publ. No. 7, pp. 1-25. Am. Inst. Biol. Sci., Washington, D.C. Knight, G. A. (1977). Factors relating to long term storage of magnetic tape. Recorded Sound 66/61, 681-692. McClurg, D. R. (1976). "Professional Recorders." Otari Corp., San Carlos, California. Olson, H. F. (1957). "Acoustical Engineering." Van Nostrand, New York.

52

David C. Wickstrom

Olson, H. F. (1972). "Modern Sound Reproduction." Van Nostrand-Reinhold, New York. Peterson, A. P. G., and Gross, E. E., Jr. (1972). "Handbook of Noise Measurement," 7th ed. General Radio, Concord, Massachusetts. Peus, S. (1977). Microphones and transients. Sound Eng. Mag. 11, 35-38. Roederer, J. G. (1975). "Introduction to the Physics and Psychophysics of Music." SpringerVerlag, Berlin and New York. Tall, J. T. (1958). "Techniques of Magnetic Recording." Macmillan, New York. Temmer, S. F. (1979). Vu meter and peak program meter, peaceful coexistence. Prepr. No. 1474(G-5). Audio Eng. Soc, New York. Toombs, D. (1981). "Sound recording." David & Charles, London and North Pomfret, Vermont. Tremaine, H. M. (1969). "Audio Cyclopedia," 2nd ed. Sams & Bobbs-Merrill, New York. Watkins, W. A. (1967). The harmonic interval; fact or artifact in spectral analysis of pulse trains. In "Marine Bioacoustics" (W. N. Tavolga, ed.), Vol. 2, pp. 15-43. Pergamon, New York. Weber, P. J. (1967). "The Tape Recorder as an Instrumentation Device." Ampex Corp., Redwood City, California. Woram, J. M. (1976). "The Recording Studio Handbook." Sagamore, Plainview, New York.